radv: allow to capture SQTT traces with RADV_THREAD_TRACE=<start_frame>

This is pretty basic (and a bit crappy at the moment). I think we
might want some sort of overlay in the future and also be able to
trigger captures with F12 or whatever.

To record a capture, set RADV_THREAD_TRACE to something greater than
zero (eg. RADV_THREAD_TRACE=100 will capture frame #100). If the
driver didn't crash (or the GPU didn't hang), the capture file
should be stored in /tmp.

To open that capture, use Radeon GPU Profiler and enjoy your
profiling times with RADV! \o/

Note that thread trace support is quite experimental, only GFX9 is
supported at the moment, and a bunch of useful stuff are still missing
(shader ISA, pipelines info, etc). More is comming soon.

Signed-off-by: Samuel Pitoiset <[email protected]>
Reviewed-by: Bas Nieuwenhuizen <[email protected]>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3900>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3900>

3 files changed, 68 insertions, 5 deletions

diff --git a/src/amd/vulkan/radv_device.c b/src/amd/vulkan/radv_device.c
index 04fdb462a92..0ffcf8bd49c 100644
--- a/src/amd/vulkan/radv_device.c
+++ b/src/amd/vulkan/radv_device.c
@@ -2969,6 +2969,24 @@ VkResult radv_CreateDevice(
 		radv_dump_enabled_options(device, stderr);
 	}
 
+	int radv_thread_trace = radv_get_int_debug_option("RADV_THREAD_TRACE", -1);
+	if (radv_thread_trace >= 0) {
+		fprintf(stderr, "*****************************************************************************\n");
+		fprintf(stderr, "* WARNING: Thread trace support is experimental and only supported on GFX9! *\n");
+		fprintf(stderr, "*****************************************************************************\n");
+
+		/* TODO: add support for more ASICs. */
+		assert(device->physical_device->rad_info.chip_class == GFX9);
+
+		/* Default buffer size set to 1MB per SE. */
+		device->thread_trace_buffer_size =
+			radv_get_int_debug_option("RADV_THREAD_TRACE_BUFFER_SIZE", 1024 * 1024);
+		device->thread_trace_start_frame = radv_thread_trace;
+
+		if (!radv_thread_trace_init(device))
+			goto fail;
+	}
+
 	/* Temporarily disable secure compile while we create meta shaders, etc */
 	uint8_t sc_threads = device->instance->num_sc_threads;
 	if (sc_threads)
@@ -3044,6 +3062,8 @@ fail_meta:
 fail:
 	radv_bo_list_finish(&device->bo_list);
 
+	radv_thread_trace_finish(device);
+
 	if (device->trace_bo)
 		device->ws->buffer_destroy(device->trace_bo);
 
@@ -3093,6 +3113,9 @@ void radv_DestroyDevice(
 
 	pthread_cond_destroy(&device->timeline_cond);
 	radv_bo_list_finish(&device->bo_list);
+
+	radv_thread_trace_finish(device);
+
 	if (radv_device_use_secure_compile(device->instance)) {
 		for (unsigned i = 0; i < device->instance->num_sc_threads; i++ ) {
 			destroy_secure_compile_device(device, i);
diff --git a/src/amd/vulkan/radv_private.h b/src/amd/vulkan/radv_private.h
index 7c4032d1ec6..1f776c8f580 100644
--- a/src/amd/vulkan/radv_private.h
+++ b/src/amd/vulkan/radv_private.h
@@ -859,6 +859,7 @@ struct radv_device {
 	struct radeon_winsys_bo *thread_trace_bo;
 	void *thread_trace_ptr;
 	uint32_t thread_trace_buffer_size;
+	int thread_trace_start_frame;
 };
 
 struct radv_device_memory {
diff --git a/src/amd/vulkan/radv_wsi.c b/src/amd/vulkan/radv_wsi.c
index a2b0afa48c3..2da2049feb4 100644
--- a/src/amd/vulkan/radv_wsi.c
+++ b/src/amd/vulkan/radv_wsi.c
@@ -269,16 +269,55 @@ VkResult radv_AcquireNextImage2KHR(
 	return result;
 }
 
+/* TODO: Improve the way to trigger capture (overlay, etc). */
+static void
+radv_handle_thread_trace(VkQueue _queue)
+{
+	RADV_FROM_HANDLE(radv_queue, queue, _queue);
+	static bool thread_trace_enabled = false;
+	static uint64_t num_frames = 0;
+
+	if (thread_trace_enabled) {
+		struct radv_thread_trace thread_trace = {};
+
+		radv_end_thread_trace(queue);
+		thread_trace_enabled = false;
+
+		/* TODO: Do something better than this whole sync. */
+		radv_QueueWaitIdle(_queue);
+
+		if (radv_get_thread_trace(queue, &thread_trace))
+			radv_dump_thread_trace(queue->device, &thread_trace);
+	} else {
+		if (num_frames == queue->device->thread_trace_start_frame) {
+			radv_begin_thread_trace(queue);
+			assert(!thread_trace_enabled);
+			thread_trace_enabled = true;
+		}
+	}
+	num_frames++;
+}
+
 VkResult radv_QueuePresentKHR(
 	VkQueue                                  _queue,
 	const VkPresentInfoKHR*                  pPresentInfo)
 {
 	RADV_FROM_HANDLE(radv_queue, queue, _queue);
-	return wsi_common_queue_present(&queue->device->physical_device->wsi_device,
-					radv_device_to_handle(queue->device),
-					_queue,
-					queue->queue_family_index,
-					pPresentInfo);
+	VkResult result;
+
+	result = wsi_common_queue_present(&queue->device->physical_device->wsi_device,
+				          radv_device_to_handle(queue->device),
+				          _queue,
+				          queue->queue_family_index,
+				          pPresentInfo);
+	if (result != VK_SUCCESS)
+		return result;
+
+	if (unlikely(queue->device->thread_trace_bo)) {
+		radv_handle_thread_trace(_queue);
+	}
+
+	return VK_SUCCESS;
 }